Skip to main content
Topic: Spectra versus XCMS/Camera (Read 3778 times) previous topic - next topic

Spectra versus XCMS/Camera

Hello,

I just wanted to double check the following approach is not incorrect. I may be overthinking, but better safe than sorry.

For metabolite identification, I am searching for the feature xcms spat out, but I predominantly search the peaks that turn up in my spectra. Occasionally, these may not perfectly match either Camera or Xcms features.
I guess the question is: what's deemed more reliable - the xcms feature detection or the spectrum in identifying molec ions for compound identification?

Many thanks!

Re: Spectra versus XCMS/Camera

Reply #1
Are you talking about which m/z value to use? The one from the peaktable or from the raw data?
In xcms the m/z for each feature in individual samples is the intensity weighted mean across the peak. Then when you group features across samples it uses the median m/z of those mean values.

Because of this averaging the m/z in your peaktable should have a bit better accuracy. That is under the assumption that the parameters were sane enough not to group things that are NOT the same compound. So using this value you should normally be able to restrict your m/z range more when you search.
Blog: stanstrup.github.io

Re: Spectra versus XCMS/Camera

Reply #2
HI Jan,
Thank you!
Yes, I mean mz values. So the peaktable mz has a better accuracy than the peak mz shown in an actual spectrum at the same retention time?

Re: Spectra versus XCMS/Camera

Reply #3
Given that the peak picking went well and didn't merge several masses it shouldn't have, yes that is my experience. If you see big differences you know something is not going right.
Blog: stanstrup.github.io

Re: Spectra versus XCMS/Camera

Reply #4
Thank you.
The problem I find most often with peak picking is that it picks random low intensity peaks within a certain retention time. And the other issue I seem to find are separated mzs, where you get essentially the same or very similar mz for the same RT.